Who gets credit? A striatal mechanism for resolving associative ambiguity

Speaker: Associate Professor Jay Bertran-Gonzalez
Director, the Australasian Course in Advanced Neuroscience
Head, Neuromodulatory Systems and Behaviour group
Decision Neuroscience Laboratory
University of New South Wales
 

Who gets credit? A striatal mechanism for resolving associative ambiguity

Learning begins in a state of ambiguity. Rewards rarely follow a single isolated action, but instead occur amidst a rich mixture of movements, exploratory behaviours and environmental events. A fundamental challenge for the brain is therefore determining which elements of experience truly predict outcomes and should be incorporated into future behaviour. In this talk, I will describe unpublished evidence for a recently described striatal cellular mechanism that contributes directly to this selection process. We found that dopamine D2 receptor-expressing neurons are transiently recruited during the earliest stages of instrumental learning, when action-outcome contingencies first emerge. These neurons exert a powerful cell-to-cell influence over neighbouring dopamine D1 receptor-expressing neurons, locally constraining dopamine-dependent plasticity despite ongoing dopamine release. Using molecular, optical and behavioural approaches, we gained real-time control over this process and found that enhancing it durably improves the encoding of action-outcome relationships under conditions of reinforcement uncertainty. Our findings suggest that the striatum does not simply encode reinforcement signals, but actively filters competing plasticity events to promote the incorporation of behaviourally relevant associations. More broadly, they identify a potential neural solution to the classic problem of credit assignment during learning.